摘要:
由于双酚A(BPA)具有内分泌干扰等毒性效应,有些国家已经开始限制其在部分工业品生产中使用。一些双酚A类似物因用于替代BPA而被大量使用。因其与双酚A具有相似的分子结构,双酚A类似物是否也具有内分泌干扰等毒性效应受到越来越多的关注。采用逐步多元线性回归(MLR)方法,构建了可预测双酚A类似物雌激素效应的定量结构-活性关系模型。模型的决定系数R2=0.899,去一法交叉验证系数Q2LOO=0.868,Bootstrapping验证系数Q2BOOT=0.755,均方根误差RMSE=0.339,表明模型具有较好的拟合优度、稳健性;验证集决定系数R2EXT=0.921,外部验证系数Q2EXT=0.810,均方根误差RMSE=0.638,表明模型具有较好的预测能力。采用欧几里德距离方法和Williams图表征了模型应用域,依据分子描述符对模型进行了机理解释,并用所建模型,填补了22种其他双酚A类似物缺失的雌激素干扰效应数据。
Abstract:
Due to the endocrine disrupting effects on organisms, bisphenol A (BPA) has been banned from certain products in some countries. Alternatively, BPA analogs (bisphenols) were introduced to replace BPA. However, owing to the structural similarity of those compounds to BPA, there is an increasing concern on the issue that whether bisphenols also could disturb the endocrine system of organisms. In the present study, stepwise multiple linear regressions (MLR) analysis was employed to select variables and develop quantitative structure-activity relationship (QSAR) models for predicting the estrogenic activity of bisphenols. The coefficient determination (R2), cross validated Q2 leave-one-out (Q2LOO), bootstrapping coefficient (Q2BOOT) and the root mean square error (RMSE) of the model is 0.899, 0.868, 0.755 and 0.339, respectively, indicating that the model has high goodness-of-fit and robustness. In addition, the coefficient determination (R2EXT), external validation coefficient (Q2EXT ) and RMSE of the validation set is 0.921, 0.810 and 0.638, respectively, implying that the model has good predictive ability. The model application domain was characterized by the Euclidean distance and Williams plot. Mechanisms interpretation of the developed model was also performed based on the selected molecular descriptors. Finally, with the help of this model, the data gap for other twenty-two bisphenols on their estrogenic activity was filled.
点击查看大图
下载:
